This post intends to take a deep look at the words for "dog" in Japanese, "inu" and "ken", both written with the same kanji (sinogram; Chinese character): 犬.
I will begin with some basic phonological and etymological information, then move to an elaboration of the immediate cause for the writing of this post, observations from colleagues, and a brief conclusion.
For those who are unfamiliar with Japanese, most kanji (Sinograms) have at least two quite different pronunciations, an on['yomi] 音[読み] (Sino-Japanese reading) and a kun['yomi] 訓[読み] (native reading). In the case of 犬 ("dog"), the former is ken and the latter is inu. They both mean "dog". The underlying question that prompted me to write this post is when to use which. And that came in the form of a long message from Nathan Hopson, which I quote here:
In recent years, Japanese television has embraced a role as cheerleader for “Cool Japan,” producing a mess of self-congratulatory paeans to Japan in the form of shows about:
a) Japanese people, culture, and businesses succeeding in the world, or
b) how much foreigners (especially, but not exclusively Westerners, i.e. gaijin) love those same Japanese people, culture, and businesses
The segment I saw recently on TV Aichi’s どうぶつピース!! (Dōbutsu Pīsu!!) about Japanese dogs shows how far this goes. Yes, the dogs of war have become the dogs of culture wars. And the dogs of failed international political manipulation, according to this BBC article about Japan’s attempt to ingratiate itself with Putin by giving him a second dog ahead of his visit and talks about finally signing a peace treaty and the Japanese pipe dream of getting Russia to “return” four northern islands.
Before going any further, it’s worth noting that, in English, the title, “Animal peace!!” would be properly punctuated, “Animals, ‘Peace!’” In other words, it’s basically, “Hey animals, say ‘Cheese!’” The subtitle is “Super-cute images one after another!” Television is learning, and plagiarizing, from YouTube.
Anyway, the segment in question was one of several on the program that’s a story rather than a glorified YouTube video. It appears to be part of a longer series of stories about the popularity of Japanese dogs in the world. Television would have its captive audiences believe that Akitas and Shibas are growing popular in the world, and there’s an implicit link to Japan’s “Gross National Cool” and the whole “Cool Japan” phenomenon. This is emphasized by the Japanese names that Western owners choose for their dogs, their stories of falling in love with Japan and/or Japanese culture (usually manga and anime), etc. This is precisely the sort of thing that’s driven me to ignore Japanese TV as much as possible recently.
But my interest was piqued when my wife mentioned that she was rather suddenly bothered by something that had been nagging at me for a while now.
When I learned Japanese twenty years ago, Akitas (秋田犬) and Shibas (しば犬) were most definitely “Akita ken” and “Shiba ken,” both using the sinicized reading (on yomi) for the character for dog (犬). But in recent years, I’ve been hearing broadcasters use the “nativized” reading (kun yomi), “inu” for both.
So when my wife mentioned that she was feeling the same uneasiness about “Akita inu,” I decided to figure out what’s going on here.
First, the English sources.
In English, things are relatively simple for the Akita, because the breed name doesn’t include either inu or ken.
The American Kennel Club, which describes the breed as “Dignified, courageous, and profoundly loyal,” officially recognized the Akita in 1972, though breed standards in Japan date to the 1930s. The AKC website adds, “The ‘inu’ that is sometimes added to the name simply means ‘dog.’” There is no mention of the “ken” reading.
As far as I can tell, The Akita Club of America does not take a stand on the issue.
One important tidbit that both of these organizations (and Wikipedia) note is that, as the Akita Club puts it, “The Akita is one of Seven Breeds designated as a National Monument in his native country of Japan.”
So on to the Japanese sources.
The Agency for Cultural Affairs, which manages such designations, is unambiguous: the reading in the Agency database (Japanese only) for Akitas is あきたいぬ (Akita inu), and “ken” is nowhere to be found. So officially, inu it is.
The website of Odate City, the dog’s ancestral home (it was previously referred to as the Odate dog and added to the list of National Monuments as the Akita in 1931), is also firm on this point.
What of other Japanese native dogs?
The Shiba is another Japanese dog popular outside Japan, and it is also referred to in the media using inu rather than ken. The Agency for Cultural Affairs agrees, though I have to admit this weirds me out a bit, too.
So what about the other five dogs registered as National Monuments? (And what about the word “monument” here…?) Well, I just about threw my hands up and walked away when I put the database results into a list:
秋田犬 （あきたいぬ） Akita
越の犬 （こしのいぬ） Koshi or Koshino (extinct)
甲斐犬 （かいけん） Kai
紀州犬 （きしゅうけん）Kishu (Kishū)
土佐犬 （とさけん）Tosa (distinct from the Tosa fighting dog, apparently)
FWIW, the Japanese dog fan community site nihonken.org (note “ken!!”) lists:
Hokkaido Inu (北海道犬)
Shiba Inu (柴犬)
Kai Ken (甲斐犬)
Kishu Ken (紀州犬)
Shikoku Ken (四国犬)
Well, as you might expect, it turns out that my wife and I are not the only Japanese speakers bothered by this. A quick Google search yields dozens of questions like this one (in Japanese) in online Q&A informational forums like Yahoo! Japan’s Chiebukuro (“Adviser”) alluding to confusion or frustration with this choice.
So why is it that the government agency in charge of cultural affairs and the media persist with the inu pronunciation?
Turns out, NHK has an answer. And yes, it’s an answer to a perplexed viewer’s question. And yes, I felt vindicated by this. But then I saw the date: 2001. So this has apparently been “a thing” for much longer than I realized. Feelings of vindication melted away….
Japan’s public broadcaster keeps a manual (ことばのハンドブック, Kotoba no handobukku) for just such questions of linguistic usage. So when a viewer / listener asked about NHK’s policy on the discrepant readings of 犬, the handbook came to the rescue with a general non-answer:
“The readings are decided based on consideration of both concerned organizations (such as local preservation societies) and customary common usage.”
But the specifics are actually quite interesting:
“For instance, in its home region, 秋田犬 is traditionally referred to as ‘Akita inu’ … However, a February 1990 poll of Tokyo male and female residents sixteen years old and older revealed that 95% pronounced this name ‘Akita ken.’ For this reason, NHK decided to add ‘Akita ken’ to the traditional ‘Akita inu’ as acceptable for use in broadcasting.”
The answer continues with a list of NHK’s accepted pronunciations, which differs from the Agency for Cultural Affairs:
しば犬 [×柴犬] Shiba
(Note that the “x” means that NHK does not accept the kanji 柴 for the dog breed, though it is used both commonly and by the Agency for Cultural Affairs)
This left me wondering about how and why “Akita ken,” and for that matter “Shiba ken,” had become so commonplace. My assumption is that it’s because of parsing issues. When treated as a suffix for the preceding word, “-ken” as in Nihonken (Japanese dog/s), it’s treated as one character of a longer Sinitic compound. When treated as a separate word, as in “Akita + dog,” then inu makes sense. However, that’s a harder sell given normal Japanese parsing patterns.
Any insights? And what about analogous examples in other languages?
Nathan's observations might very well stand as a substantial Language Log guest post on their own, and indeed they are, but I'm adding a few things below to supplement all the good information that he has provided.
The pronunciation of 犬 (犭when used as a radical on the left side of a character) in MSM is quǎn, in Cantonese is hyun2, in Hakka is khién / khián, in Southern Min is khián, in Wu is qyoe2, in Middle Sinitic is /kʰwenX/, and in Old Sinitic is /*[k]ʷʰˤ[e][n]ʔ (Baxter-Sagart) or /*kʰʷeːnʔ/ (Zhengzhang).
(There's a quite different word for "dog" in Chinese, viz., gǒu 狗, but I will leave that for another occasion, just as I will not discuss the two main words for dog in English, "dog" [PIE root unknown] and "hound" [has a PIE root], though others may wish to say something about the Chinese and English pairs in the comments.)
Thus we know very well where the character 犬 and the pronunciation "ken" come from, but where does "inu" come from? I have the same sort of question about all Japanese pairs of on and kun readings, so it will be a treat for me to look intensively at this one particular case.
I'm the proud owner of Yamanaka Jōta 山中襄太, Kokugo gogen jiten 国語語源辞典 (Etymological Dictionary of the National Language) (Tokyo: Azekura Shobō 校倉書房, 1976, 1993 [4th ed.]). On p. 78a, the author lists the following words for "dog" in Tungusic topolects of Manchuria: ina, inau, inai, inaki, inda, and nenda.
I cannot help but think that these Tungusic words and Japanese "inu" are somehow related. Roy Andrew Miller famously believed that Japanese and Korean were members of the Altaic language group. The consensus view of professional linguists, however, is that there's not even a well-established Altaic language group, much less one that includes Korean and Japanese. See J. Marshall Unger, "Summary report of the Altaic panel", in Philip Baldi, ed., Linguistic Change and Reconstruction Methodology (Berlin: Mouten de Gruyter, 1990), pp. 479-482.
The Manchu word for "dog" is "indahun". I have a hunch that "hun" is some sort of suffix. See the remarks of Juha Janhunen and Jichang Lulu below.
Comments by historical linguists and philologists
At first I thought that this should be an easy answer: in compounds in which the character 犬 follows a name, it should generally be properly inu, since names would tend to be kunyomi-based, but readers tend to find it difficult to decide whether a reading of a name is kunyomi (calling for the kunyomi inu) or onyomi (calling for the onyomi ken). The Nihon kokugo daijiten from Shôgakukan lists only Shibainu for 柴犬. It favors Akitainu, although it also lists Akitaken as an alternate for 秋田犬. For 日本犬, Nihonken is the main entry, although Nipponinu is an alternate (I'm glad we're not touching on Nihon vs. Nippon). I wasn't really aware of a chronology for the bifurcation, but it would seem to be recent.
NHK has an answer to the query 'is it inu or ken?' online. They say that in their practice of the moment, some compounds are only read XXinu (e.g. Shibainu), some are only read XXken (e.g. Kaiken, Kishûken, Hokkaidôken, Karafutoken), and some are read either way (e.g., Akitainu/Akitaken, Tosainu/Tosaken). They explain that the official name of the Akita breed is Akitainu, according to the locally-based preservation society (Akitainu hozonkai), but a 1990 survey of males and females over sixteen in Tokyo found that 95% say "Akitaken." So the choice depends on the context. https://www.nhk.or.jp/bunken/summary/kotoba/gimon/079.html
I have heard that the Nihonken hozon-kai 日本犬保存会 (Japan Dog Preservation Society) changed its reading from inu to ken twenty-five years ago, but I cannot confirm. If I had to guess, I would place the change about then.
On the face of it, it seems like a typical on/kun division of labor to me. Yamato words are not good for compounding with proper names, so parallel to koinu 子犬 ("puppy") vs. Akita-ken 秋田犬 you get koushi 子牛 but Wagyū 和牛.
In Modern Japanese the normal word is native Japanese inu. Sino-Japanese ken occurs mostly in idiomatic expressions like ken-en no naka 犬猿の仲: the relationship between a dog and a monkey, which is an idiom for the relationship between irreconcilable enemies.
Well, ken is the usual Sino-J reading for 犬 and inu is its usual J gloss. Both denote 'dog', but the connotations of inu used as a free noun differ from its use in compounds (inu by itself can mean 'spy' too). Ken cannot be used as a free noun; it only occurs in compounds.
Sino-Japanese ken is today normally used in the sense of dog breed, e.g. ainu ken 'Ainu dog' or akita ken 'Akita dog'. Inu is the regular word when used alone, or also in some fixed expressions like koma inu 'Korean dog' = the Chinese mythical dog-lion hybrid. The word inu has no generally accepted etymology but has been compared with Tungusic nginakin 'dog'. It can hardly be connected with the Eurasian kyon-kywen > ken-quan 犬 etymon, which is also present in Korean gae and Ghilyak kan/ng. [VHM: Korean gae reminds me of Sinitic gǒu 狗.] Manchu indahun corresponds exactly to Ewenki nginakin. The Proto-Tungusic shape would have been *ngïnda-kun, with *-kun (> Ewenki -kin) apparently as a suffix but with no specifically identifiable meaning – perhaps diminutive.
Indahūn (from older indahon; Jurchen maybe *indahu) is Manchu for 'dog', and the -hV(n) part would look like some sort of suffix since elsewhere in Tungusic there are dog words without it (Oroch inda). But also with something resembling it (something like ninakin in Evenki), meaning the suffix isn't necessarily exclusive to Manchu. The elephant in the room is of course the Japanese and there have been attempts to link these Tungusic dog words to it.
Pamela Kyle Crossley:
Manchu for dog is “indahûn”. Vaguely like Japanese inu, huh? As for other Tungusic languages, I don’t know. Manchu is certainly not borrowed from Turkic or Mongolian, which have words like “ köpek” and "nokhoi" for dog. Daniel Kane (Kitan language and script) thought Kitan might have been ni.qo (Liáo shǐ guóyǔ jiě 遼史國語解 ["Explanation of the National Language in the Official Dynastic History of the Liao Dynasty]) gives the Chinese transcription as niehe) or something close, and gives “it” as Old Turkic.
Daniel Kane (responding to Pamela Crossley's note on ni.go being the Kitan word for "dog"):
Yes, more or less, but like most Kitan words, there is a sort of fuzziness as one tries to balance the Kitan script version, the Chinese transcription (Liaoshi [official history of the Liao Dynasty] usually) and various forms of Mongol. In this case we are lucky we have several sources, but still they do not jell exactly. I think ni-qo is as good as any – about 70% likely = which in Kitan is pretty high!
Here is a group of entries from the Sìtǐ hébì wénjiàn 四體合壁文鑒 (Four-script combined textual mirror printed during the Qing / Manchu Dynasty (provided courtesy of Pamela Crossley):
The entries in this section all mean what the Chinese means:
Chinese: wán yīng quǎn lèi 頑鷹犬類 ("playing with falcons / eagles and dogs"), where wán 頑 ("obstinate: stubborn; recalcitrant") = 玩 ("play; enjoy; have fun")
Tibetan: khra khyi rtse ba'i skor ("about / concerning / pertaining to the play / games / enjoyment of hawks and dogs") — hunting with falcons and dogs
khra=hawk, khyi=dog, rtse ( or brtse)=play/game, skor= about/around
Manchu: giyahvn indahvn efire hacin ("the category of sporting with falcons and dogs")
Mongolian: qarcaghai noqai naghadqu jüil ("section / part [about] playing [with] hawks [and] dogs")
The Manchu to the right of the Chinese gives Manchu phonetic glosses for the Chinese phrase.
Two final notes on the history of the word quǎn / inu / ken 犬 ("dog") in East Asia:
- I could not find this kanji / hanzi in John R. Bentley, ABC Dictionary of Ancient Chinese Phonograms (Honolulu: University of Hawaii Press, 2016).
- Axel Schuessler, in his ABC Etymological Dictionary of Old Chinese (Honolulu: University of Hawaii Press, 2007), p. 437, notes that this word survives in Min topolects, but has been replaced by gǒu 狗 in most of the others.
[Thanks to Steve Wadley, Leonard W.J. van der Kuijp, Douglas Duckworth, Elliot Sperling, Matthew Kapstein, Nathan Hill, Gray Tuttle, Bob Ramsey, and David Prager Branner]